We ensure that the sampling function is explicitly called during the image generation process after obtaining z_mean and z_log_var.

Generate different images each time by explicitly calling the sampling function with new random noise.

import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import matplotlib.pyplot as plt

The process of sampling the latent vector involves generating a random sample from the distribution defined by 𝑧 mean and z log_var. This is done using the reparameterization trick, which allows gradients to flow through the sampling process during training.

# Sampling function with randomness
def sampling(z_mean, z_log_var):
    batch = tf.shape(z_mean)[0]
    dim = tf.shape(z_mean)[1]
    epsilon = tf.keras.backend.random_normal(shape=(batch, dim))
    return z_mean + tf.exp(0.5 * z_log_var) * epsilon

# VAE Class with custom call method
class VAE(keras.Model):
    def __init__(self, encoder, decoder, **kwargs):
        super(VAE, self).__init__(**kwargs)
        self.encoder = encoder
        self.decoder = decoder

    def call(self, inputs):
        z_mean, z_log_var, z = self.encoder(inputs)
        reconstructed = self.decoder(z)
        reconstruction_loss = tf.reduce_mean(
            tf.keras.losses.binary_crossentropy(inputs, reconstructed)
        )
        reconstruction_loss *= 200 * 200 * 3
        kl_loss = 1 + z_log_var - tf.square(z_mean) - tf.exp(z_log_var)
        kl_loss = tf.reduce_mean(kl_loss)
        kl_loss *= -0.5
        self.add_loss(reconstruction_loss + kl_loss)
        return reconstructed

# Encoder
latent_dim = 2

encoder_inputs = keras.Input(shape=(200, 200, 3))
x = layers.Conv2D(32, 3, activation="relu", strides=2, padding="same")(encoder_inputs)
x = layers.Conv2D(64, 3, activation="relu", strides=2, padding="same")(x)
x = layers.Flatten()(x)
x = layers.Dense(16, activation="relu")(x)
z_mean = layers.Dense(latent_dim, name="z_mean")(x)
z_log_var = layers.Dense(latent_dim, name="z_log_var")(x)
z = layers.Lambda(lambda args: sampling(*args), output_shape=(latent_dim,), name='z')([z_mean, z_log_var])

encoder = keras.Model(encoder_inputs, [z_mean, z_log_var, z], name='encoder')
encoder.summary()

Model: "encoder"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                  ┃ Output Shape              ┃         Param # ┃ Connected to               ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ input_layer (InputLayer)      │ (None, 200, 200, 3)       │               0 │ -                          │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ conv2d (Conv2D)               │ (None, 100, 100, 32)      │             896 │ input_layer[0][0]          │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ conv2d_1 (Conv2D)             │ (None, 50, 50, 64)        │          18,496 │ conv2d[0][0]               │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ flatten (Flatten)             │ (None, 160000)            │               0 │ conv2d_1[0][0]             │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ dense (Dense)                 │ (None, 16)                │       2,560,016 │ flatten[0][0]              │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z_mean (Dense)                │ (None, 2)                 │              34 │ dense[0][0]                │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z_log_var (Dense)             │ (None, 2)                 │              34 │ dense[0][0]                │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z (Lambda)                    │ (None, 2)                 │               0 │ z_mean[0][0],              │
│                               │                           │                 │ z_log_var[0][0]            │
└───────────────────────────────┴───────────────────────────┴─────────────────┴────────────────────────────┘

 Total params: 2,579,476 (9.84 MB)

 Trainable params: 2,579,476 (9.84 MB)

 Non-trainable params: 0 (0.00 B)

# Decoder
latent_inputs = keras.Input(shape=(latent_dim,))
x = layers.Dense(50 * 50 * 64, activation="relu")(latent_inputs)
x = layers.Reshape((50, 50, 64))(x)
x = layers.Conv2DTranspose(64, 3, activation="relu", strides=2, padding="same")(x)
x = layers.Conv2DTranspose(32, 3, activation="relu", strides=2, padding="same")(x)
decoder_outputs = layers.Conv2DTranspose(3, 3, activation="sigmoid", padding="same")(x)

decoder = keras.Model(latent_inputs, decoder_outputs, name='decoder')
decoder.summary()

Model: "decoder"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ input_layer_1 (InputLayer)           │ (None, 2)                   │               0 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ dense_1 (Dense)                      │ (None, 160000)              │         480,000 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ reshape (Reshape)                    │ (None, 50, 50, 64)          │               0 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose (Conv2DTranspose)   │ (None, 100, 100, 64)        │          36,928 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose_1 (Conv2DTranspose) │ (None, 200, 200, 32)        │          18,464 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose_2 (Conv2DTranspose) │ (None, 200, 200, 3)         │             867 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 536,259 (2.05 MB)

 Trainable params: 536,259 (2.05 MB)

 Non-trainable params: 0 (0.00 B)

# VAE Model
vae = VAE(encoder, decoder)
vae.compile(optimizer='adam')
vae.build(input_shape=(None, 200, 200, 3))
vae.summary()

# Prepare and normalize the image
pic_2 = keras.preprocessing.image.load_img('pic_2.jpeg', target_size=(200, 200))
pic_2 = keras.preprocessing.image.img_to_array(pic_2).astype("float32") / 255
pic_2 = np.expand_dims(pic_2, 0)

# Train the VAE model
vae.fit(pic_2, epochs=100, batch_size=1)

Model: "vae"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ encoder (Functional)                 │ ?                           │       2,579,476 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ decoder (Functional)                 │ ?                           │         536,259 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 3,115,735 (11.89 MB)

 Trainable params: 3,115,735 (11.89 MB)

 Non-trainable params: 0 (0.00 B)

Epoch 1/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 2s 2s/step - loss: 83178.2500
Epoch 2/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 54ms/step - loss: 83129.2500
Epoch 3/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 83552.8594
Epoch 4/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 82888.7812
Epoch 5/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 83003.5703
Epoch 6/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 82873.5703
Epoch 7/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 80328.5781
Epoch 8/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 103015.8984
Epoch 9/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 81020.2734
Epoch 10/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 82548.5391
Epoch 11/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 82767.2734
Epoch 12/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 82772.3750
Epoch 13/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 82759.4844
Epoch 14/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 82740.7656
Epoch 15/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 82719.9766
Epoch 16/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 82767.0391
Epoch 17/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 82705.4688
Epoch 18/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 82680.0156
Epoch 19/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 82687.8828
Epoch 20/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 82682.8281
Epoch 21/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 82598.8672
Epoch 22/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 114ms/step - loss: 82637.7812
Epoch 23/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 82592.1875
Epoch 24/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 82544.9453
Epoch 25/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 54ms/step - loss: 82512.3828
Epoch 26/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 59ms/step - loss: 82422.1797
Epoch 27/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 82315.6719
Epoch 28/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 59ms/step - loss: 82401.1406
Epoch 29/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 82375.1953
Epoch 30/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 82416.9453
Epoch 31/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 82308.6953
Epoch 32/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 82082.1953
Epoch 33/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 82227.6172
Epoch 34/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 82238.7031
Epoch 35/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 81953.7969
Epoch 36/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 82037.8750
Epoch 37/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 82018.2031
Epoch 38/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 82014.2812
Epoch 39/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 81661.3438
Epoch 40/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 80818.4844
Epoch 41/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 82053.0547
Epoch 42/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 79494.8359
Epoch 43/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 80898.1875
Epoch 44/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 79759.9375
Epoch 45/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 79260.4375
Epoch 46/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 59ms/step - loss: 76992.2734
Epoch 47/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 79651.1875
Epoch 48/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 81640.4766
Epoch 49/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 81179.8125
Epoch 50/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 80421.4844
Epoch 51/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 90ms/step - loss: 79459.7109
Epoch 52/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 79605.2266
Epoch 53/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 77276.2812
Epoch 54/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 76496.4766
Epoch 55/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 80ms/step - loss: 76753.7344
Epoch 56/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 74851.1719
Epoch 57/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 77927.4219
Epoch 58/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 74561.1562
Epoch 59/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 111720.9141
Epoch 60/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 73512.0859
Epoch 61/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 76679.8359
Epoch 62/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 76298.5703
Epoch 63/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 78089.7812
Epoch 64/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 78323.0625
Epoch 65/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 79091.9688
Epoch 66/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 79524.1953
Epoch 67/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 54ms/step - loss: 78808.1172
Epoch 68/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 77712.7109
Epoch 69/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 77794.5547
Epoch 70/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 78462.0469
Epoch 71/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 57ms/step - loss: 77828.1641
Epoch 72/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 52ms/step - loss: 77288.6953
Epoch 73/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 76320.1953
Epoch 74/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 76509.0156
Epoch 75/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 75829.4219
Epoch 76/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 74ms/step - loss: 74081.6094
Epoch 77/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 75310.8984
Epoch 78/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 70599.2656
Epoch 79/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 69802.5703
Epoch 80/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 86031.2734
Epoch 81/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 68547.1094
Epoch 82/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 71655.6562
Epoch 83/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 74415.7734
Epoch 84/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 72088.7812
Epoch 85/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 58ms/step - loss: 71316.7969
Epoch 86/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 86ms/step - loss: 72163.7656
Epoch 87/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 113ms/step - loss: 71978.0234
Epoch 88/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 71232.8516
Epoch 89/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 75ms/step - loss: 71490.9062
Epoch 90/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 54ms/step - loss: 71052.8516
Epoch 91/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 70762.4375
Epoch 92/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 70316.2109
Epoch 93/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step - loss: 71599.3281
Epoch 94/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 68455.8125
Epoch 95/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 56ms/step - loss: 70260.8125
Epoch 96/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 53ms/step - loss: 68871.9062
Epoch 97/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 69570.7812
Epoch 98/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 67728.8359
Epoch 99/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 55ms/step - loss: 67470.1953
Epoch 100/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 66914.5156

<keras.src.callbacks.history.History at 0x2a770907f10>

# Show plot and save image function
def show_and_save_plot(image, save_path):
    plt.imshow(image.squeeze())
    plt.axis('off')
    plt.savefig(save_path)
    plt.show()


for i in range(10):
    z_mean, z_log_var, _ = vae.encoder.predict(pic_2)
    # Introduce randomness in sampling
    encoded_imgs = sampling(z_mean, z_log_var).numpy()
    decoded_imgs = vae.decoder.predict(encoded_imgs)
    save_path = f'generated_image_{i + 1}.png'
    show_and_save_plot(decoded_imgs, save_path)

print("Images have been saved and displayed successfully.")

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 86ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 60ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 35ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 35ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 37ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 35ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 20ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 22ms/step

Images have been saved and displayed successfully.

What output do you see? Explain why

For the codes above, the resulted 10 images vary much better than using pic_1

z_mean, z_log_var, _ = vae.encoder.predict(pic_2)
print("z_mean:", z_mean)
print("z_log_var:", z_log_var)

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 23ms/step
z_mean: [[-0.8666203  -0.60099775]]
z_log_var: [[0.26988217 0.63260573]]

I still want to try latent_dim = 10

import numpy as np
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
import matplotlib.pyplot as plt

# Sampling function with randomness
def sampling(z_mean, z_log_var):
    batch = tf.shape(z_mean)[0]
    dim = tf.shape(z_mean)[1]
    epsilon = tf.keras.backend.random_normal(shape=(batch, dim))
    return z_mean + tf.exp(0.5 * z_log_var) * epsilon

# VAE Class with custom call method
class VAE(keras.Model):
    def __init__(self, encoder, decoder, **kwargs):
        super(VAE, self).__init__(**kwargs)
        self.encoder = encoder
        self.decoder = decoder
        self.mse = tf.keras.losses.MeanSquaredError()

    def call(self, inputs):
        z_mean, z_log_var, z = self.encoder(inputs)
        reconstructed = self.decoder(z)
        reconstruction_loss = tf.reduce_mean(
            self.mse(inputs, reconstructed)
        )
        reconstruction_loss *= 200 * 200 * 3
        kl_loss = 1 + z_log_var - tf.square(z_mean) - tf.exp(z_log_var)
        kl_loss = tf.reduce_mean(kl_loss)
        kl_loss *= -0.5
        self.add_loss(reconstruction_loss + kl_loss)
        return reconstructed

# Encoder
latent_dim = 10  # Increased latent dimension

encoder_inputs = keras.Input(shape=(200, 200, 3))
x = layers.Conv2D(32, 3, activation="relu", strides=2, padding="same")(encoder_inputs)
x = layers.Conv2D(64, 3, activation="relu", strides=2, padding="same")(x)
x = layers.Flatten()(x)
x = layers.Dense(16, activation="relu")(x)
z_mean = layers.Dense(latent_dim, name="z_mean")(x)
z_log_var = layers.Dense(latent_dim, name="z_log_var")(x)
z = layers.Lambda(lambda args: sampling(*args), output_shape=(latent_dim,), name='z')([z_mean, z_log_var])

encoder = keras.Model(encoder_inputs, [z_mean, z_log_var, z], name='encoder')
encoder.summary()

# Decoder
latent_inputs = keras.Input(shape=(latent_dim,))
x = layers.Dense(50 * 50 * 64, activation="relu")(latent_inputs)
x = layers.Reshape((50, 50, 64))(x)
x = layers.Conv2DTranspose(64, 3, activation="relu", strides=2, padding="same")(x)
x = layers.Conv2DTranspose(32, 3, activation="relu", strides=2, padding="same")(x)
decoder_outputs = layers.Conv2DTranspose(3, 3, activation="sigmoid", padding="same")(x)

decoder = keras.Model(latent_inputs, decoder_outputs, name='decoder')
decoder.summary()

# VAE Model
vae = VAE(encoder, decoder)
vae.compile(optimizer='adam')
vae.build(input_shape=(None, 200, 200, 3))
vae.summary()

# Prepare and normalize the image
pic_2 = keras.preprocessing.image.load_img('pic_2.jpeg', target_size=(200, 200))
pic_2 = keras.preprocessing.image.img_to_array(pic_2).astype("float32") / 255
pic_2 = np.expand_dims(pic_2, 0)

# Train the VAE model
vae.fit(pic_2, epochs=100, batch_size=1)

# Show plot and save image function
def show_and_save_plot(image, save_path):
    plt.imshow(image.squeeze())
    plt.axis('off')
    plt.savefig(save_path)
    plt.show()

# Run the encoder-decoder multiple times to generate different copies
for i in range(10):
    z_mean, z_log_var, _ = vae.encoder.predict(pic_2)
    # Introduce randomness in sampling
    encoded_imgs = sampling(z_mean, z_log_var).numpy()
    decoded_imgs = vae.decoder.predict(encoded_imgs)
    save_path = f'generated_image_{i + 1}.png'
    show_and_save_plot(decoded_imgs, save_path)

print("Images have been saved and displayed successfully.")

Model: "encoder"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                  ┃ Output Shape              ┃         Param # ┃ Connected to               ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ input_layer_2 (InputLayer)    │ (None, 200, 200, 3)       │               0 │ -                          │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ conv2d_2 (Conv2D)             │ (None, 100, 100, 32)      │             896 │ input_layer_2[0][0]        │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ conv2d_3 (Conv2D)             │ (None, 50, 50, 64)        │          18,496 │ conv2d_2[0][0]             │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ flatten_1 (Flatten)           │ (None, 160000)            │               0 │ conv2d_3[0][0]             │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ dense_2 (Dense)               │ (None, 16)                │       2,560,016 │ flatten_1[0][0]            │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z_mean (Dense)                │ (None, 10)                │             170 │ dense_2[0][0]              │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z_log_var (Dense)             │ (None, 10)                │             170 │ dense_2[0][0]              │
├───────────────────────────────┼───────────────────────────┼─────────────────┼────────────────────────────┤
│ z (Lambda)                    │ (None, 10)                │               0 │ z_mean[0][0],              │
│                               │                           │                 │ z_log_var[0][0]            │
└───────────────────────────────┴───────────────────────────┴─────────────────┴────────────────────────────┘

 Total params: 2,579,748 (9.84 MB)

 Trainable params: 2,579,748 (9.84 MB)

 Non-trainable params: 0 (0.00 B)

Model: "decoder"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ input_layer_3 (InputLayer)           │ (None, 10)                  │               0 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ dense_3 (Dense)                      │ (None, 160000)              │       1,760,000 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ reshape_1 (Reshape)                  │ (None, 50, 50, 64)          │               0 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose_3 (Conv2DTranspose) │ (None, 100, 100, 64)        │          36,928 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose_4 (Conv2DTranspose) │ (None, 200, 200, 32)        │          18,464 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ conv2d_transpose_5 (Conv2DTranspose) │ (None, 200, 200, 3)         │             867 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 1,816,259 (6.93 MB)

 Trainable params: 1,816,259 (6.93 MB)

 Non-trainable params: 0 (0.00 B)

Model: "vae_1"

┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Layer (type)                         ┃ Output Shape                ┃         Param # ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ encoder (Functional)                 │ ?                           │       2,579,748 │
├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤
│ decoder (Functional)                 │ ?                           │       1,816,259 │
└──────────────────────────────────────┴─────────────────────────────┴─────────────────┘

 Total params: 4,396,007 (16.77 MB)

 Trainable params: 4,396,007 (16.77 MB)

 Non-trainable params: 0 (0.00 B)

Epoch 1/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 2s 2s/step - loss: 9969.8330
Epoch 2/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 9952.2275
Epoch 3/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 9905.1992
Epoch 4/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 9799.8447
Epoch 5/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 9443.6328
Epoch 6/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 8943.6104
Epoch 7/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 70ms/step - loss: 8196.6943
Epoch 8/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 6990.1729
Epoch 9/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 6634.2773
Epoch 10/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 6583.5356
Epoch 11/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 5908.4053
Epoch 12/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 5622.7407
Epoch 13/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 76ms/step - loss: 5539.8311
Epoch 14/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 76ms/step - loss: 5298.5249
Epoch 15/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 78ms/step - loss: 5238.3037
Epoch 16/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 70ms/step - loss: 5181.6914
Epoch 17/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 5127.2827
Epoch 18/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 5079.2339
Epoch 19/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 5033.1089
Epoch 20/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 72ms/step - loss: 4981.9697
Epoch 21/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 73ms/step - loss: 4940.1826
Epoch 22/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 116ms/step - loss: 4901.8281
Epoch 23/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 85ms/step - loss: 4861.6597
Epoch 24/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 74ms/step - loss: 4831.6338
Epoch 25/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 74ms/step - loss: 4815.4307
Epoch 26/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 73ms/step - loss: 4781.6743
Epoch 27/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 73ms/step - loss: 4777.3623
Epoch 28/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 4735.8325
Epoch 29/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 4715.5664
Epoch 30/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 4692.4648
Epoch 31/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 4653.4199
Epoch 32/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 4624.4077
Epoch 33/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 4687.2280
Epoch 34/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 4562.8481
Epoch 35/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 4552.0024
Epoch 36/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 4529.4150
Epoch 37/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 4474.5405
Epoch 38/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 4438.9492
Epoch 39/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 80ms/step - loss: 4424.4653
Epoch 40/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 76ms/step - loss: 4372.9590
Epoch 41/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 4338.4277
Epoch 42/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 4306.4731
Epoch 43/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 4260.6147
Epoch 44/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 4218.7573
Epoch 45/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 4180.4497
Epoch 46/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 4124.2217
Epoch 47/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 4065.3379
Epoch 48/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 4011.7961
Epoch 49/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 3945.3098
Epoch 50/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 3866.5088
Epoch 51/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 3787.2458
Epoch 52/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 70ms/step - loss: 3693.9219
Epoch 53/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 3578.2249
Epoch 54/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 3462.9343
Epoch 55/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 73ms/step - loss: 3297.0491
Epoch 56/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 79ms/step - loss: 3129.5481
Epoch 57/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 2919.8857
Epoch 58/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 77ms/step - loss: 2685.6443
Epoch 59/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 2456.9216
Epoch 60/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 72ms/step - loss: 2239.6851
Epoch 61/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 2009.8368
Epoch 62/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 1831.4513
Epoch 63/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 71ms/step - loss: 1681.6642
Epoch 64/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 1522.0482
Epoch 65/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 1371.6166
Epoch 66/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 1306.6349
Epoch 67/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 1223.9176
Epoch 68/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 1108.3302
Epoch 69/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 1037.7194
Epoch 70/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 64ms/step - loss: 983.1470
Epoch 71/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 947.4807
Epoch 72/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 70ms/step - loss: 909.9151
Epoch 73/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 873.0816
Epoch 74/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 72ms/step - loss: 835.9836
Epoch 75/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 74ms/step - loss: 871.8846
Epoch 76/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 73ms/step - loss: 799.4135
Epoch 77/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 67ms/step - loss: 785.8033
Epoch 78/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 738.4094
Epoch 79/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 736.1783
Epoch 80/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 106ms/step - loss: 706.0345
Epoch 81/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 74ms/step - loss: 696.6925
Epoch 82/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 72ms/step - loss: 672.9388
Epoch 83/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 662.8980
Epoch 84/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 70ms/step - loss: 646.0948
Epoch 85/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 635.3652
Epoch 86/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 625.7963
Epoch 87/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 620.7348
Epoch 88/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 607.3733
Epoch 89/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 595.3042
Epoch 90/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 75ms/step - loss: 597.9349
Epoch 91/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 579.2706
Epoch 92/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 615.9193
Epoch 93/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 61ms/step - loss: 569.0283
Epoch 94/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 69ms/step - loss: 642.7772
Epoch 95/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 68ms/step - loss: 555.5855
Epoch 96/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 609.1636
Epoch 97/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 65ms/step - loss: 546.4082
Epoch 98/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 66ms/step - loss: 685.6516
Epoch 99/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 63ms/step - loss: 538.1050
Epoch 100/100
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 62ms/step - loss: 602.5829
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 82ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 59ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 20ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 17ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 17ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 18ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 17ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 31ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 18ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 21ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 18ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 19ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 25ms/step

Images have been saved and displayed successfully.